1. Field of the Invention
The present invention relates to a monitoring system and method, information processing device, and program, and more particularly to a monitoring system and method, information processing device, and program which decrease the processing load of a server related to a collation processing system implemented by a server which performs collation processing between facial images captured by a plurality of monitoring cameras and facial images in a database, and improve the collation accuracy.
2. Description of the Related Art
The technology for executing collation processing based on facial images by storing facial images in a database, determining similarity with a captured facial image and searching facial images of which similarity is higher than a predetermined value, and managing data based on the facial images using the collation result, is becoming popular.
As a technology for managing data based on facial images, a technology for improving collation accuracy using a first server and a second server, for example, has been proposed (Japanese Patent Application Laid-Open No. 2001-187249).
If the above mentioned method, which aims to improve collation accuracy, is used, however, and if terminal devices are installed at each game machine of a game center, and facial images are acquired and supplied to the server, a large volume of facial image data is supplied from terminal devices to the server, the server accesses the facial image database and repeats collation processing between an enormous volume of facial images and the database, so the processing load of the server increases, and as a result collating processing time elapses and real-time processing is affected.
For example, in the case of a game center having 1000 game machine units, where 20% of players are “switching machines”, “temporarily leaving”, “coming” and “leaving” within one hour, collation processing is required for about 200 players. In other words, collation processing for 2400 players is required if the game center operates 12 hours a day. For “switching game machines”, “coming” and “leaving”, collation must be performed on the server to manage customers by facial images, but for “temporarily leaving”, as in going to the restroom, purchasing food and drink, eating and making phone calls, it is unnecessary to supply data on facial images to the server and perform collation, since the same player returns.
Also in collation, similarity of a facial image registered in a database and a facial image of an individual is provided as a score, and if the score exceeds a threshold, the “individual” is identified as the actual person, but facial images which can easily be used for facial image collation, in terms of “direction of face (up, down)”, “inclination of face” and “brightness of face” must be searched and registered in a database.
However in the current situation, facial images in a facial image database are not updated to optimum facial images, and any images that are captured as facial images are registered as the facial images of an individual.
Also a system of managing data based on facial images is comprised of a terminal device for capturing facial images and a server, but if communication with a server is interrupted for a predetermined period, the terminal device must either discard the captured facial images, or otherwise continue storing facial images without end, which may crash the system.
In any case, in the above mentioned data management system which has a database of facial images, processing greatly depends on the server which executes collation processing, which could increase load and decrease real-time processing and accuracy of collation processing based on facial images.